Optical coherence tomography (OCT) is an optical signal acquisition and processing method. It captures micrometer resolution 3D-images within optically scattering media, such as biological tissue. A distance from a scattering medium is measured by use of interferometric signals. In OCT, comparatively long wavelength light, typically in the near-infrared region is used. This has the advantage that the light may penetrate rather deeply into the scattering medium, allowing to obtain sub-surface images at near-microscopic resolution. OCT has proven particularly useful in the imaging of the human eye, allowing obtaining high resolution images of the retina and the anterior segment of the eye. A further very attractive application of OCT currently under development is intravascular imaging in cardiology.
As in any interferometric methods, it is possible to measure the interference signal in the time domain or in the frequency domain. Simply put, in the time domain the length of the interferometer reference arm is varied and the intensity of the interference signal is measured, without paying attention to the signal spectrum. Alternatively, the interference of the individual spectral components can be measured, which corresponds to the measurement in the frequency domain. There are also mixed variants of time and frequency domain OCT, of which the “time encoded frequency domain OCT”, which is also referred to as the “swept source OCT” has recently received particular attention. In the swept source OCT, each frequency scan allows to obtain exactly one depth scan of the OCT image. There is a common understanding that the swept source OCT is the most powerful and promising OCT variant that will be of increasing importance, in particular for ophthalmologic and cardiologic OCT imaging. In order to acquire the largest number of images in a given time, it is, however, necessary to provide a tunable light source that allows a fast frequency sweep.
Unfortunately, conventional tunable lasers have a fundamental limitation in the sweep rate, because for every resonance frequency, the stimulated emission of the laser medium has to be built up again, see e.g. Huber, R., et al., “Amplified, frequency swept lasers for frequency domain reflectometry and OCT imaging: design and scaling principles”. Optics Express, 2005. 13(9): p. 3513-3528. In fact, the fastest frequency sweeps of conventional lasers are at about 200 kHz, and for fundamental reasons there is little hope to push this limit much higher.
Currently, there are three promising approaches to obtain tunable laser sources allowing for higher sweep rates:    (i) The Fourier domain mode locking (FDML) techniques as described in Huber, R., M. Wojtkowski, and J. G. Fujimoto, “Fourier Domain Mode Locking (FDML): A new laser operating regime and applications for optical coherence tomography”. Optics Express, 2006. 14(8): p. 3225-3237. An FDML laser typically comprises a semiconductor optical amplifier as a light source, a fiber Fabry-Pérot filter acting as an optical band pass filter and a delay line that is adapted such that the light circulation time is an integer multiple of the filter sweep frequency. A portion of the circulating laser light is then extracted by a fiber coupler. The idea underlying the FDML technique is that light passing through the filter takes just as long to propagate through the laser cavity and arrive back at the filter as the filter takes for sweeping the frequency once (or an integer multiple times). This way, wave trains with different wavelengths can pass through the gain medium of the laser several times and are thereby amplified, since any part of the wave train from the previous roundtrip acts as a seed for stimulated emission of the gain medium with the resonator tuned to the current wavelength.    (ii) A wavelength swept amplified spontaneous emission source as described in Eigenwillig, C. M., et al., Wavelength swept amplified spontaneous emission source. Optics Express, 2009. 17(21): p. 18794-18807. In a wavelength swept amplified spontaneous emission source, amplified spontaneous emission light alternately passes a cascade of optical gain elements and tunable band pass filters.    (iii) Tunable vertical-cavity surface emitting lasers (VCSELs). A dynamically tunable VCSEL has recently been proposed in Jayaraman, V., et al. “OCT Imaging up to 760 kHz Axial Scan Rate Using Single-Mode 1310 nm MEMS-Tunable VCSELs with >100 nm Tuning Range” in CLEO. 2011. The VCSEL of Jayaraman et al. comprises a fully oxidized AlGaAs mirror to which an InP-based multi-quantum well is wafer bonded, the multi-quantum well representing an active medium. A dielectric mirror is provided opposite the multi-quantum well, with an air gap in between. Accordingly, the dielectric mirror and the fully oxidized AlGaAs mirror form a resonator cavity and hence effectively a Fabry-Pérot filter. The top dielectric mirror is flexibly suspended by rather delicate material bridges and can be moved up and down by electrostatic actuation, thereby forming a Fabry-Pérot tunable filter device. Note in this regard that in the present disclosure, the term “Fabry-Pérot filter device” shall be given the broadest possible meaning and shall in particular not be limited to passive filters. The tunable VCSEL is a micro-electromechanical system (MEMS) device based on semiconductor fabrication techniques, i.e. deposition of material layers, patterning by photolithography and etching to produce the required shapes.
Accordingly, each of the three abovementioned light sources with a potential for higher frequency sweep rates than conventional tunable lasers employs some sort of Fabry-Pérot filter device, and the sweep frequency of the Fabry-Pérot filter device turns out to be the limiting factor of the sweep frequency of the light source as a whole.
FIG. 1 is a schematic view of a fiber Fabry-Pérot tunable filter (FFP-TF) 10 as known from prior art, see for example U.S. Pat. No. 6,904,206 B2 or U.S. Pat. No. 5,208,886. The FFP-TF 10 comprises two mirrors 12 deposited directly to the ends of optical fibers 14 which are accommodated in ferrules 16 held in holding pieces 18 which are in turn connected via a piezo-electric actuator 20.
Between the two mirrors 12, an air gap is formed that defines an optical resonator cavity. By operation of the piezo-actuator 20, the optical path length between the two mirrors 12, or in other words, the length of the resonator cavity formed between the two mirrors 12 can be tuned. When the optical length of the round trip length of the cavity is an integer of a wavelength, then this wavelength together with a narrow band resonates inside the cavity and passes through the filter 10 with very low loss. Wavelengths not meeting this resonance condition, however, will not pass from one optical fiber 14 to the other, but will be blocked by the filter instead. This way, the FFP-TF 10 acts as tunable narrow band pass filter.
The FFP-TF 10 design of FIG. 1 has proven to be very robust and reliable in operation. However, it turns out that the tuning rate, i.e. the sweep frequency of the FFP-TF 10 is limited. For example, a FFP-TF of a design as shown in FIG. 1 that is commercially available from Micron Optics, Inc. allows for a tuning frequency of up to 50 kHz only.
A further FFP-TF is commercially available from LambdaQuest, see Lambda Quest, L. “High Speed Optical Tunable Filter 2011”; Available from: http://www.lambdaquest.com/products.htm. The LambdaQuest FFP-TF has a U-shaped design, where the two arms of the U carry the ferrules with the fibers and where the arms are connected by a piezo-actuator. The general shape is shown in FIG. 2 in a simplified way, where the ferrules, the mirrors and the fibers have been omitted. Instead, FIG. 2 only shows the two arms 22 that in the actual device would carry the ferrules and the fibers, which are connected by a piezo-actuator 24.
At low operation frequencies of the piezo-actuator 24, i.e. at low sweep frequencies, the arms 22 will simply be moved to and fro according to the expansion and contraction of the piezo-actuator 24. At higher frequencies, the entire U-shaped structure will start to oscillate in a vibration mode that is indicated by the arrows 26 in FIG. 2, where the arrows 26 resemble the (heavily exaggerated) local oscillation amplitude.
The vibration mode of the U-shaped structure can also be seen in the response function of the LambdaQuest filter recorded by the inventors which is shown in FIG. 3. The response function characterizes the modulation amplitude of the resonator cavity as a function of driving frequency of the piezo-actuator 24. As is seen in FIG. 3, this response function increases from some low frequency value, also referred to “DC value” herein, to some peak a little over 50 kHz, where the modulation amplitude of the air gap is the largest. Beyond this peak, the response function drops below the DC value. In practice, this means that beyond the resonance peak, the amplitude of the air gap modulation will be generally too small to allow for the suitable tunable frequency range. In fact, the highest sweep frequency of the LambdaQuest filter according to the manufacturer is 40 kHz.
Both FFP-TFs of FIGS. 1 and 2 are ordinary “mechanical” devices, where mechanical components like the holding pieces 18 and the ferrules 16 of FIG. 1 or the arms 22 of FIG. 2 are mechanically moved by a piezo-actuator 20, 24, respectively. It is currently found difficult to push the sweep frequency limit significantly higher than the 40 or 50 kHz of the fastest currently commercially available FFP-TFs when employing an ordinary mechanical design as shown in FIGS. 1 and 2. Instead, for higher scanning frequencies, MEMS devices are often believed to be more promising. A good example is the MEMS-based VCSEL of Jayaraman et al. discussed above, where the suspended top dielectric mirror can be moved at a frequency of as high as 380 kHz.
However, VCSELS based on MEMS technology also have considerable drawbacks in practice. The manufacturing requires a large number of etching steps that need to be carried out in a clean room. Also, since the material bridges suspending the mirror are very thin and hence difficult to manufacture, there is an inherent reproducibility problem. Also, since the top mirror is very thin, heat dissipation is rather difficult. Further, since the top mirror is actuated electrostatically, the necessary control voltages are rather high and the elongation depends highly non-linearly on the control voltage. And while in general the extremely small mass of the top mirror is an advantage when it comes to sweep frequency, the resonance wavelength tends to be unstable due to thermal noise (Brownian motion), see J Dellunde, et al., “Trans-verse-mode selection and noise properties of external-cavity vertical-cavity surface-emitting lasers including multiple-reflection effects”, Journal of the Optical Society of America B-Optical Physics, vol. 16, pp. 2131-2139, November 1999.